Small-molecule targeted therapies have demonstrated outstanding potential in the clinic. These drugs are designed to minimize adverse effects by selectively attacking cancer cells while exerting minimal damage to normal cells. Although initial response to targeted therapies may be high, yielding positive response rates and often improving survival for an important percentage of patients, resistance often limits long-term effectiveness. On the other hand, immunotherapy has demonstrated durable results, yet for a limited number of patients. Growing evidence indicates that some targeted agents can modulate different components of the antitumor immune response. These include immune sensitization by inhibiting tumor cell–intrinsic immune evasion programs or enhancing antigenicity, as well as direct effects on immune effector and immunosuppressive cells. The combination of these two approaches, therefore, has the potential to result in synergistic and durable outcomes for patients. In this review, we focus on the latest advances on integrating immunotherapy with small-molecule targeted inhibitors. In particular, we discuss how specific oncogenic events differentially affect immune response, and the implications of these findings on the rational design of effective combinations of immunotherapy and targeted therapies.
Over the past decade, immunotherapy has cemented its status as a vital component of cancer care. In particular, immune checkpoint blockade (ICB) agents, most notably inhibitors of the PD-1/PD-L1 and CTLA-4 pathways, have become standard-of-care for many solid and hematologic malignancies, leading to durable results and improved long-term protection from relapse. This latter effect is likely mediated by the induction of an adaptive immune memory capable of eradicating otherwise obstinate tumor cells. Despite their broad applicability, however, ICB benefits only a limited number of patients. Best durable responses have been observed in melanoma, where 5-year survival was reported at 26% for ipilimumab (anti-CTLA-4) and 44% for nivolumab (anti-PD-1), and non–small cell lung cancer (NSCLC), where overall survival approximates 16% after 5 years (1, 2). In an effort to harness this unique potential, the research field has seen a revamped focus on understanding how current and new therapies can influence antitumor immune response. In fact, multiple strategies aiming to potentiate immunotherapy are currently under preclinical and clinical investigation. Conventional cancer therapies and small-molecule targeted inhibitors have been shown to modulate various components of tumor immunity and response to immunotherapy (3–5). Targeted agents, in particular, exert these effects by altering mechanisms of immune escape encoded by oncogenic pathways in tumor cells. Therefore, a deeper understanding of how specific oncogenic events shape tumor immunity will prove crucial to the successful development of immunomodulatory strategies. To this end, targeted therapies are ideally situated to block or enhance relevant pathways while exerting minimal damage to normal cells.
Immune Evasion by Cancer Cells
The process by which tumor cells evade immune surveillance can be better understood through the concept of immunoediting (1, 6). Initially, malignant cells are regularly detected and eliminated by the immune system through recognition of immunogenic antigens and generation of an innate and adaptive immune response. Acute inflammation activates innate immunity, leading to dendritic cell (DC) maturation and subsequent priming of T cells, which are central to the antitumor response. This constant pressure, however, may act to select for tumor cells that are able to escape immune attack and remain in equilibrium until further changes promote overt tumor growth. This final process is usually accompanied by a shift from acute to chronic inflammation and the establishment of an immunosuppressive tumor microenvironment (TME) via recruitment of immune suppressive cells whose normal function is to dampen immune response, including regulatory T cells (TRegs), protumorigenic tumor-associated macrophages (TAM) and myeloid-derived suppressor cells (MDSC), which further facilitate tumor growth. As a consequence, T cells in general become exhausted or dysfunctional, and therefore unable to mount an effective antitumor response (Fig. 1).
Oncogenic Signaling Pathways Shape the Tumor Immune Microenvironment
Oncogenic signaling pathways have the potential to affect every component of tumor immunity. Careful analysis of clinical studies and the development of relevant animal models are key steps to maximize translational potential. Studies using genetically engineered mouse models (GEMM) correlated with clinical data have provided much insight into how specific oncogenic events differentially contribute to immune escape. Mechanistic studies for target prediction and biomarker discovery, as well as preclinical evaluation in mouse models thus provide important information for designing potentially successful clinical trials (Fig. 2). Importantly, mechanisms of immune evasion and de novo as well as acquired resistance to immunotherapy often overlap, thus underscoring the potential for targeted approaches that could simultaneously sensitize tumors to immunotherapy and prevent recurrence. In this section, we review some of the major molecular mechanisms described to date.
The MYC oncogene was shown to directly upregulate expression of the innate immune inhibitor receptor CD47, a so-called “don't eat me” signal, and of the adaptive immune checkpoint ligand PD-L1 in lymphoma/leukemia models of conditional MYC overexpression (Fig. 3A; ref. 7). These results were subsequently corroborated by multiple groups in different cancer models (8). Furthermore, conditional MYC activation in a KRASG12D-driven model of lung cancer showed that MYC drives tumor progression and recruitment of an immunosuppressive TME characterized by a marked influx of macrophages and depletion of T cells, B cells, and natural killer (NK) cells (9). These effects were mediated by tumor-secreted CCL9 and IL23, which enhanced recruitment of PD-L1+ macrophages and promoted lymphocyte exclusion, respectively (Fig. 3A; ref. 9). In turn, MYC deactivation reversed these changes and led to tumor regression, which was dependent on NK cells but not on T cells (9). Notably, CCL9/IL23 co-blockade inhibited tumor progression, while PD-L1 blockade restored T-cell infiltration but did not measurably affect tumor growth (9). More recently, newly developed small-molecule MYC inhibitors that disrupt MYC/MAX dimerization were shown to promote antitumor immune response, and to synergistically inhibit tumor growth of MyC-Cap mouse prostate cancer allografts when combined with PD-1 blockade (10).
KRASG12D was shown to mediate immune suppression in a GEMM of colorectal carcinoma with inducible KRASG12D and additional APC and p53 double deletion (11). In this case, KRASG12D repressed expression of IRF2, thus alleviating repression of CXCL3 expression by colorectal carcinoma tumor cells and promoting recruitment of CXCR2+ MDSCs to the TME (Fig. 3B; ref. 11). While single agents against PD-1 or CXCR2 did not affect tumor growth or survival, combined treatment significantly increased survival and inhibited tumor growth (11). Furthermore, a novel KRASG12C-specific inhibitor, AMG510, strongly promoted a proinflammatory TME and synergized with anti-PD-1 to inhibit mouse syngeneic CT-26 colorectal carcinoma tumors with enforced KRASG12C expression, which led to complete regression in 90% of cases (9/10) and immunologic memory, as shown by the ability to reject a second challenge of CT-26 tumor cells (12).
EGFR and HER2
Mutant EGFR in lung cancer mouse models has been shown to promote the establishment of an immunosuppressive TME characterized by low levels of cytotoxic T lymphocytes (CTL) and increased markers of T-cell exhaustion (Fig. 3C; ref. 13). Ectopic mutant EGFR expression in bronchial epithelial BEAS2B cells upregulates PD-L1 expression, while small-molecule EGFR inhibition in NSCLC cell lines downregulates PD-L1 (13). Consistently, mouse lung adenocarcinoma tumors driven by EgfrL858R display high myeloid cells infiltration, reduced CD4+ T-helper response, and blunted CD8+ T-cell expansion, compared with tumors driven by KrasG12D or concomitant KrasG12D and p53 deletion (14). In the case of HER2, HER2-positive breast cancers predominantly exhibit immune subtypes consistent with ongoing immune activity, including IFNγ-dominant phenotype (∼50% of cases; characterized by strong CD8+ and antitumorigenic macrophage signals) and wound-healing phenotype (∼44% of cases; characterized by high expression of angiogenic genes, high proliferation, and TH2-type responses; refs. 15, 16). In addition to inhibiting oncogenic HER2 signaling in tumor cells, anti-HER2–targeted mAbs stimulate innate and adaptive immune responses critical for clinical efficacy (17). These effects are mediated primarily via antibody-dependent cell-mediated cytotoxicity, antibody-dependent cellular phagocytosis, and by inducing antigen cross-presentation and T-cell priming (17). Considering the aggressive nature of HER2+ breast cancers and the outstanding therapeutic effect of anti-HER2 mAbs, these observations underscore the power of the immune system to subdue highly malignant tumor cells.
Genetic loss of PTEN is associated with reduced antitumor immunity in multiple cancers (18–20). In melanoma, PTEN deficiency correlated with decreased response to ICB in a cohort of patients (n = 39), and with decreased immune activation scores in melanoma samples from The Cancer Genome Atlas (TCGA; ref. 20). Interestingly, PTEN deficiency and WNT/β-catenin pathway activation were largely nonoverlapping (20). Using a BRAF-mutant melanoma xenograft model with ectopic expression of melanoma antigen gp100 and MHC class I H2-Db, which is specifically recognized by CD8+ T cells from transgenic PMEL-1 mice, it was shown that PTEN silencing in tumor cells reduced T-cell infiltration and cytotoxic response (Fig. 3D; ref. 20). Moreover, because PTEN-deficient tumors preferentially signal through PI3Kβ (21), treatment with the PI3Kβ isoform-specific inhibitor GSK2636771 improved response to PD-1 blockade in a GEMM of BRAFV600E/PTEN-null melanoma (20). Similarly, a novel chimeric GEMM of metastatic castration-resistant prostate cancer (mCRPC) with triple deletion of PTEN, p53, and Smad4 showed markedly enhanced response to combined PD-1/CTLA-4 blockade when combined with GSK2636771 (22). These mCRPC tumors were highly infiltrated by Gr-MDSCs, which contributed to primary resistance to immunotherapy, and showed synergistic response to ICB in combination with targeted agents that preferentially affect Gr-MDSCs, such as the tyrosine kinase inhibitor cabozantinib, the PI3K/mTOR dual inhibitor BEZ235 and the CXCR1/2 inhibitor SX-682 (22). Of note, the same group had previously reported that additional loss of Smad4 in a PTEN-null prostate cancer GEMM dramatically enhances tumor progression, metastatic spread, and lethality (23), and upregulates CXCL5 expression in tumor cells via HIPPO-YAP1 signaling, which enhances recruitment of immune suppressive CXCR2+ MDSCs (22).
Analysis of human melanoma samples revealed a correlation between T-cell exclusion and WNT/β-catenin signaling, including gain-of-function mutations on the β-catenin gene (CTNNB1) and upregulated expression of β-catenin target genes (24). To further investigate these findings, the authors compared a GEMM of metastatic melanoma driven by BRAFV600E and PTEN loss with a syngeneic model harboring additional constitutively active β-catenin, thus showing that β-catenin inhibits production of CCL4 by tumors cells, which leads to impaired recruitment of CD103+ DCs and consequently to impaired T-cell activation (Fig. 3E; ref. 24). Of note, while BRAFV600E/PTEN-null tumors responded to combined PD-L1/CTLA-4 blockade and exhibited significant growth inhibition, tumors with additional β-catenin activation failed to respond to this immunotherapy (24). Consistently, WNT/β-catenin signaling was found to inversely correlate with T-cell infiltration in colorectal cancer (25), and across multiple cancer types compiled from TCGA (26).
Loss of LKB1 in a mouse model of NSCLC driven by mutant KRAS results in neutrophil accumulation and increased T-cell exhaustion (27). Interestingly, LKB1 loss is associated with decreased PD-L1 expression and resistance to PD-1 blockade in mouse models and patient tumors (27). Indeed, retrospective analyses of clinical response in patients with KRAS-mutant lung adenocarcinoma identified genomic mutations on LKB1 as a significant biomarker for primary resistance to anti-PD-1/PD-L1 immunotherapy, as well as in another cohort of NSCLC irrespective of KRAS status (28). Further work demonstrated that LKB1 deficiency in KRAS-mutant lung cancer results in downregulation of STING and, consequently, an inability to respond to cytoplasmic double-stranded DNA (dsDNA; ref. 29). STING downregulation facilitates immune escape by preventing STING-mediated expression of type I IFNs and proinflammatory cytokines, which are necessary for proper engagement and activation of antitumor immune response (Fig. 3F; ref. 30).
STAT3 and NFκB
Signaling pathways that regulate expression of inflammatory cytokines, such as STAT3 and NFκB, have the potential to dramatically affect immune response. STAT3 can promote immune escape by upregulating immune suppressive genes, including IL6, IL10, TGFβ, and VEGF, while simultaneously downregulating immune effector genes such as type I and II IFNs, IL12, CD80, CD86, MHC class II molecules, CCL5, and CXCL10 (31). Tumor cell–intrinsic STAT3 promotes paracrine activation of STAT3 in various populations of immune cells, thereby reducing NK and T-cell cytotoxicity, inhibiting DC maturation and TH1-type response, and stimulating immunosuppressive cells such as MDSCs, TRegs, and TAMs (Fig. 3G; refs. 32, 33). In a breast cancer GEMM driven by the polyoma virus middle T antigen (PyMT), which is characterized by aggressive and metastatic tumors with latencies around 3 to 4 weeks and 80% penetrance, genetic ablation of Stat3 resulted in early hyperplastic lesions that were readily cleared by the immune system, although after a latency averaging 40 weeks, 30% of these mice developed nonmetastatic tumors that escaped immune surveillance and markedly lacked immune infiltration (34). In addition, STAT3 inhibits expression of numerous immunostimulatory genes downstream of NFκB (31). The NFκB pathway plays an important role in activating programs of immune response; however, aberrant NFκB signaling has been shown to exert strong oncogenic effects by upregulating genes that promote cell proliferation and survival (35). STAT3 binding to NFκB promotes transactivation of oncogenic genes and prevents binding to genes involved in immune response (31, 36). Furthermore, multiple upstream events, including growth factor and cytokine receptors, nonreceptor tyrosine kinases like Src and Abl, and Toll-like receptors induce STAT3 and NFκB activation either directly or indirectly via autocrine and paracrine signaling (31).
Focal adhesion kinase (FAK) was shown to induce CD8+ T-cell exhaustion and promote TReg recruitment via regulation of multiple cytokines, including CCL1/5/7, CXCL10, and TGFβ2, in a mouse model of squamous cell carcinoma (Fig. 3I), and these effects could be reversed by pharmacologic targeting of FAK by VS-4718 (38). Similar findings were described in pancreatic ductal adenocarcinoma (PDAC), where FAK inhibition with VS-4718 renders KrasG12D; Trp53L/+ PDAC tumors sensitive to adoptive cell transfer (ACT) or PD-1 blockade immunotherapy (39).
Integrating Small-molecule Targeted Therapy and Immunotherapy to Improve Therapeutic Outcomes
Distinct small-molecule targeted therapies have been shown to exert specific effects on antitumor immune response in mouse models and in the clinic (Fig. 4A). Inhibitors of BRAF, cyclin-dependent kinase 4 and 6 (CDK4/6) and PARP 1/2 are currently being tested in combination with ICB in clinical trials and have thus far shown promising potential. In this section, we discuss these three kinds of inhibitors as examples of targeted agents with immune modulatory properties.
Treatment with BRAF inhibitors has been shown to increase melanoma differentiation antigen (MDA) expression and presentation by tumor cells, increase NK-cell infiltration, and reduce TReg and MDSC levels in cell and mouse models of BRAF-mutant melanoma (Fig. 4B; refs. 40–42). Using the SM1 model of BRAFV600E mouse melanoma and SM1 cells stably expressing the chicken ovalbumin (OVA) antigen (SM1-OVA), treatment with the BRAF inhibitor vemurafenib improved ACT immunotherapy with T cells specific against OVA as well as with PMLE-1 T cells recognizing endogenous gp100 in SM1 cells (43). Furthermore, BRAF inhibition with dabrafenib in combination with the MEK inhibitor trametinib enhanced PMLE-1 ACT, leading to increased CD8+ T-cell infiltration and cytotoxicity, and complete tumor regressions (44). Combined dabrafenib and trametinib also improved response to PD-1 blockade in this model (44). Analysis of biopsy samples from patients with metastatic melanoma also revealed an association between treatment with combined BRAF and MEK inhibition, and increased MDA expression and CD8+ T-cell infiltration (45). More recently, results from a randomized phase II clinical trial of combined dabrafenib, trametinib, and PD-1 blockade by pembrolizumab compared with dabrafenib, trametinib, and placebo showed encouraging results, including improved progression-free survival and enhanced response, although the triple combination also resulted in increased adverse effects (46, 47).
CDK4/6 inhibitors exert direct immune-stimulatory effects on both tumor and immune cells (Fig. 4C). In tumor cells, the CDK4/6 inhibitors palbociclib and abemaciclib were shown to downregulate expression of the DNA methyltransferase DNMT1, leading to decreased methylation and subsequently increased expression of endogenous retrovirus elements, thus stimulating production of type III IFNs, and a consequent increase in antigen presentation and enhanced CD8+ T-cell effector function (48). Moreover, CDK4/6 inhibition specifically inhibited ex vivo proliferation of CD4+ CD25+ TRegs, but did not affect proliferation of CD4+, CD25−, and CD8+ T cells (48). Splenic CD4+ FOXP3+ TReg levels were also decreased upon treatment in vivo independently of the presence of a tumor (48). PD-L1 inhibition significantly improved response to abemaciclib in an inducible GEMM of HER2+ breast cancer, and resulted in complete tumor regression of CT-26 CRC tumors in all cases, as well as the ability to reject a second challenge with CT-26 tumor cells (48). In addition, an in vitro small-molecule screen identified CDK4/6 inhibitors as capable of directly enhancing T-cell activation via upregulation of NFAT signaling, a family of transcription factors that are required for proper activation and function of T cells (49). Consistently, CDK4/6 inhibition by palbociclib or trilaciclib potentiated PD-1 blockade to stimulate antitumor T-cell function and inhibit tumor growth in the MC38 and CT-26 CRC models (49). Interestingly, cyclin D-CDK4 was shown to promote PD-L1 proteasomal degradation (50). In vivo treatment with CDK4/6 inhibitors increased tumor PD-L1 levels and sensitized CT-26 tumors to ICB, resulting in complete tumor regression in 67% (8/12) of mice receiving combined palbociclib and anti-PD-1 (50). A study of 348 ER+/HER2− tumor samples collected from patients prior to start of CDK4/6 inhibitor treatment with palbociclib, ribociclib or abemaciclib revealed FAT1 deletion as a mechanism of therapeutic resistance (51). Mechanistically, FAT1 loss resulted in engagement of the Hippo pathway, leading to YAP/TAZ translocation to the nucleus and upregulation of CDK6 expression (51). In the clinic, preliminary results from a phase Ib clinical trial of combined abemaciclib and pembrolizumab in ER+/HER2− metastatic breast cancer have shown safety profiles similar to single agents and an initial objective response rate (ORR) of 14.3% (52).
Recent studies have demonstrated that, in addition to direct cytotoxicity, the therapeutic efficacy of PARP inhibitors (PARPi) requires coordinated activation of robust local and systemic antitumor immune response, such as increased infiltration of effector CD4+ and CD8+ T cells into the TME, increased intratumoral DCs with potent antigen-presentation capacity, and systemic reduction of MDSCs in tumor, spleen, and blood (53, 54). Mechanistically, dsDNA derived from homologous recombination (HR)-deficient tumor cells upon PARP inhibition activates cGAS/STING in tumor cells and/or DCs to drive a cGAS/STING-dependent type I IFN signal that mediates antitumor immunity (Fig. 4D; ref. 53). This mechanism of PARPi-triggered STING-dependent antitumor immunity has been demonstrated in several cancer types, including ovarian cancer, triple-negative breast cancer, and lung cancer (53–57). Interestingly, PARPis have also been shown to induce expression of PD-L1 in tumor cells via multiple mechanisms, including as a response to IFN expression, inactivation of GSK3β, reduced poly(ADP-ribosyl)ation with concomitantly increased phosphorylation of STAT3, and STING activation (56, 58–62). While PARPi-mediated PD-L1 upregulation can promote adaptative immune suppression, it can be overcome by ICB. Indeed, preclinical studies have shown that PD-1/PD-L1 blockade further augments PARPi-triggered immune response, leading to more durable suppression of tumor growth and prolonged survival (53–56). Combined PARP inhibition and ICB is being evaluated by numerous clinical trials in first-line, maintenance, and recurrent settings of both HR-deficient and HR-proficient solid tumors (63–68). In general, these trials have found combination therapies are well-tolerated, with safety concerns consistent with individual agent profiles, and have produced encouraging initial results. While PARP inhibition and PD-1/PD-L1 monotherapy exhibit low efficacy for patients with platinum-resistant ovarian cancer who lack a BRCA mutation, with ORRs approximately 5% and 4%–10%, respectively (69–74), in the ongoing phase I/II TOPACIO/KEYNOTE-162 trial, combined niraparib plus pembrolizumab demonstrated improved efficacy (ORR, 19%) in BRCA wild type patients with recurrent platinum-resistant ovarian cancer (75).
It is clear that tumor cell–intrinsic signaling mechanisms strongly affect immune composition and function. A deeper understanding of these molecular and cellular mechanisms will not only help in the design of potentially promising clinical trials of combination therapies targeted to specific groups of patients, but will also help discover new therapeutic targets with previously unknown functions in tumor immunity. Nevertheless, clinical development may still be limited by lack of significant benefit and compounding adverse effects. Careful preclinical and clinical studies are needed to improve the efficacy and tolerability of targeted therapy and immunotherapy combinations. Some areas of focus should include the need to: (i) better understand tissue-specific oncogene-related immune effects; (ii) identify and validate biomarkers to predict response and resistance to oncogene targeting; (iii) develop high fidelity animal models incorporating patient-derived tumors and humanized immune systems to better identify effective combinations without causing increased toxicity to patients; and (iv) use multiplexed assays to integrate immune and tumor intrinsic molecular changes in response to combination therapy. Still, current evidence from preclinical and clinical trials is in aggregate promising and encouraging. The notion that specific targeted agents can sensitize tumor cells to immunotherapy, thereby leading to durable and effective responses in patients that would otherwise not respond is worth pursuing. Continued basic and preclinical research integrated with careful clinical trial planning of combination therapies will likely continue to yield meaningful treatment options for patients afflicted by cancer.
Disclosure of Potential Conflicts of Interest
J.S. Bergholz reports grants from Susan G. Komen Foundation and from Friends of Dana-Farber Cancer Institute, and other from Dale Family Foundation (charitable contributions) during the conduct of the study; as well as personal fees from Geode Therapeutics (scientific consulting) outside the submitted work and a patent issued for DFCI 2180.001 (DFS-166.25). Q. Wang reports grants from Cancer Research Institute (CRI Irvington Postdoctoral Fellowship) during the conduct of the study as well as other from Crimson Biopharm (consultant) outside the submitted work and a patent issued for DFCI 2409.001 (DFS-203.60). J.J. Zhao reports grants from NIH, DoD, and Breast Cancer Research Foundation during the conduct of the study, as well as a patent pending for DFCI 2409.001 (DFS-203.60) and is a founder and director of Crimson Biopharm and Geode Therapeutics. None of the aforementioned patents are licensed to any companies. No potential conflicts of interest were disclosed by the other author.
We thank Drs. Harvey Cantor and Hye-Jung Kim for scientific discussions. We thank Elizabeth Cahn and the Breast Cancer Advocacy Group at Dana-Farber/Harvard Cancer Center for discussions on patient issues and needs. We thank the Dale Family Foundation for their charitable contributions. This work was supported in part by grants from The Susan G. Komen Foundation PDF16376814 (to J.S. Bergholz), Friends of Dana-Farber Cancer Institute (to J.S. Bergholz), Cancer Research Institute (to Q. Wang), Terri Brodeur Breast Cancer Foundation (to S. Kabraji), Breast Cancer Research Foundation (to J.J. Zhao), DoD CDMRP BC171657 (to J.J. Zhao), and NIH P50 CA168504 (to J.J. Zhao), P50 CA165962 (to J.J. Zhao), and R35 CA210057 (to J.J. Zhao).